Machine vision applied to highly variable objects

Abstract

The work presented in this paper addresses Machine Vision (MV) applications to objects of ill defined shape, size, colour or texture. Development of Machine Vision through the years from its origin in the 1950's to date is analysed and their progress and trends are discussed. The paper concentrate in the small number of MV applications to object of high variability, it discusses and present solutions to main problems associated with the MV application to this type of objects. The problem of autofocus and problems associated with changes in illumination are discussed in details and solutions presented. The paper present and discuss general software architecture for MV applications. Two examples of this type of applications are presented: detection of microorganism in water and detection of suspicious activity in humans. © 2011 Springer-Verlag.